Avicenna on Knowledge (ʿILM), Certainty (Yaqīn), Cause (ʿILLA/SABAB) and the Relative (MUḌĀF)¹

1 This study was conducted for the project Major issues and controversies of Arabic logic and philosophy of language, run by Ruhr-Universität Bochum and the University of Cambridge with the financial support of the DFG and the AHRC. I wish to thank T. Street and two anonymous referees for their comments on an earlier draft of this paper. I am responsible for any shortcomings.

Abstract

In his Kitāb al-Burhān (Book of Demonstration), Avicenna discusses a theoretical framework broadly inspired by Aristotle's Posterior Analytics which brings together logic, epistemology and metaphysics. One of the central questions explored in the book is the problem of the relation between knowledge, certainty and causal explanation. Burhān 1.8, in particular, is devoted to the analysis of how certainty comes about in causal as opposed to non-causal contexts. The distinction is understood in Avicenna's system as one between cases in which the conclusion of an argument is warranted only in virtue of an appropriate middle term, and cases in which there is no such intermediary because the predicative link between subject and predicate of the conclusion is immediate. In this context, Avicenna makes use of the case of relative terms (muḍāfāt) to clarify certain crucial aspects of his theory. The paper explores this discussion and shows how Avicenna's account of the relatively marginal role of relatives in the context of demonstration depends on insights that are central to his metaphysics and epistemology.

Keywords:

• Avicenna
• relative
• certainty
• cause
• demonstration

Notes

1 This study was conducted for the project Major issues and controversies of Arabic logic and philosophy of language, run by Ruhr-Universität Bochum and the University of Cambridge with the financial support of the DFG and the AHRC. I wish to thank T. Street and two anonymous referees for their comments on an earlier draft of this paper. I am responsible for any shortcomings.

²All translations are mine unless otherwise noted. The Arabic text of the Burhān passages discussed in the paper is that of Avicenna, Al-Šifāʾ, Al-Manṭiq, Al-Burhān (hereafter Burhān), with occasional emendations and removal of punctuation. The above passage is also quoted and translated in McGinnis, ‘Avicenna's Naturalized Epistemology and Scientific Method’, 135, which offers a useful contextualization of the problem.

واعلم أن توسط المضاف أمر قليل الجدوى في العلوم وذلك لأن نفس علمك أن زيدا أخ هو علمك بأن له أخا أو يشتمل على علمك بذلك فلا تكون النتيجة فيه شيئا أعرف من المقدمة الصغرى فإن لم يكن كذلك بل بحيث يجهل إلى أن يتبين أن له أخا فما تصورت نفس قولك زيد أخ وأمثال هذه الأشياء الأوْلى ألا تسمى قياسات فضلا عن أن تكون براهين .

³On a larger scale, the extent and importance of the study of inferences involving relational expressions for the development of Arabic logic is documented in detail by El-Rouayheb, Relational Syllogisms and the History of Arabic Logic, 900–1900.

⁴The structure of Burhān 1.8 is complex and it will be useful to offer at least a tentative map for orientation: [1] Causal certainty and why-demonstration (85.1–86.16); [2] Non-causal certainty and that-demonstration; metaphysical and epistemic immediacy (86.17–87.17); [3] Inference from effect to cause (87.18–89.3); [4] The relative case: objection and solution (89.3–90.7); [5] Repetitive syllogism and reductio (90.8–18); [6] Further discussions of causality (90.19–92.5).

⁵ Burhān 1.7 covers the distinction between that- and why-demonstration in general terms, whereas Burhān 3.3 deals more specifically with the materials discussed by Aristotle in An. Post. A13.

⁶It will be perhaps useful to note – incidentally – that a scientific theory, on Avicenna's model, will straightforwardly consist for the most part of claims that do indeed require causal justification. Immediate predications, however, are important limit-cases, as they express primitive relations within a given domain of scientific discourse with respect to which they will count as principles. The relationship between that-questions and why-questions in Avicenna's theory of science, with a specific focus on Burhān 4.1, is discussed in Strobino, ‘What If That (Is) Why? Avicenna’s Taxonomy of Scientific Inquiries’.

⁷The reason of the failure of the modal condition is that when there is an intrinsic natural ordering of properties among which asymmetrical explanatory relationships hold, no argument that fails adequately to reflect such relationships can produce certain conclusions (see in particular Burhān 1.8, 85.8). Besides, Avicenna's commitment to the view that what is caused is known with certainty only through its cause is also systematically connected to his metaphysical necessitarianism and the doctrine that whatever is necessary through another (and merely possible in itself) becomes necessary only in virtue of its cause.

⁸ فبين أن الشيء أو الحال إذا كان له سبب لم يتيقن إلا من سببه .‏

⁹فإن كان الأكبر للأصغر لا بسبب بل لذاته لكنه ليس بيّن الوجود له والأوسط كذلك للأصغر إلا أنه بيّن الوجود للأصغر ثم الأكبر بين الوجود للأوسط فينعقد برهان يقيني ويكون برهان إن ليس برهان لم وإنما كان يقينا لأن المقدمتين كليتان واجبتان ليس فيهما شك [ … ] وأما هاهنا فكان بدل السبب الذات وكان الأكبر للأصغر لذاته ولكن كان خفيا وكان الأوسط أيضاً له لذاته لا بسبب حتى إن جُهل جهل ولكنه لم يكن خفيا فقد علمت المقدمة الصغرى بوجوبها والكبرى أيضاً كذلك إذ لم يكن الأكبر للموصوفات بالأوسط إلا لذاتها لا لسبب يجهل حكمه لجهله .

¹⁰The distinction between the metaphysical level and the epistemic level is captured by the contrast between (i) belonging to something in virtue of the thing itself or in virtue of the thing's essence (li-ḏātihi) – as opposed in turn to belonging to it in virtue of a cause (li-sabab) – and (ii) belonging to something evidently (i.e. being bayyin al-wuğūd). More generally, there is a likely connection between Avicenna's concerns in Burhān 1.8 and his treatment of implicates (lawāzim) in section 1.12 of the logic of the Išārāt, which is also relevant for the commentary tradition it generates in the thirteenth century.

¹¹والذي يبقى هاهنا شيء واحد هو أن لقائل أن يقول كيف تكون الذات الواحدة تقتضي لذاتها شيئين مثلا الأصغر كيف يقتضي ب الأوسط أ الأكبر اللهم إلا أن يقتضي أحدهما لذاته أولا ويقتضي الثاني لا لذاته بل بتوسط ذلك الأول بينهما فحينئذ يكون ب علة أ لا بحسب البيان فقط بل وبحسب الوجود.‏

¹²فالجواب أن المنطقي من حيث هو منطقي يجب أن يأخذ أن هذا يمكن في مواد هذه صفتها ولا يمكن في مواد مخالفة لها وأما هل لهذه المواد إمكان أم لا وهل هذا الشك صحيح فيها أم لا فليس هو بعلم منطقي بل البحث عن أمثال هذه للفلسفة الأولى فإنه متعلق بالبحث عن أحوال الموجودات وهناك يتبين أنه يجوز أن يكون للذات الواحدة من الذوات التي ليست بغاية البساطة لواحق كثيرة تلحق معا ليس بعضها قبل بعض وأن في بعض الذوات البسيطة أحوالا تشبه هذا من جهة تركيب معنوي فيها إذ لا تكون بساطتها بساطة مطلقة وأكثر الموجودات هذه صورتها فقد تحصّل من هذا أن برهان الإن قد يعطي في مواضع يقينا دائما وأما فيما له سبب فلا يعطي اليقين الدائم بل فيما لا سبب له .

¹³I assume the simultaneity and priority implied in this passage by ‘simultaneously’ (maʿan) and ‘prior’ (qabla) to be essential, not temporal. For the distinction between temporal and essential simultaneity, priority and posteriority, see Avicenna, Ilāhiyyāt 4.1 (cf. also Ilāhiyyāt 6.2, where the discussion is articulated in terms of the relation between cause and effect).

¹⁴The view is usually associated with the idea that the middle term should be the definition or part of the definition of the major and perform an explanatory role in a demonstrative argument. See Aristotle, An. Post. B8–10 and the analysis offered in the corresponding chapters of Themistius' paraphrasis, and the commentary traditionally attributed to Philoponus. Cf. also Al-Fārābī, Kitāb al-Burhān, Chapter 2.7 (Fī l-barāhīn wa-l-asbāb), 42–44. On the relationship between definition and demonstration in Avicenna's Burhān, see Strobino, ‘Avicenna on the Indemonstrability of Definition’.

¹⁵An intriguing, although far from obvious (let alone provable) connection could be to a passage in Al-Kindī, Risāla fī waḥdāniyya Allah wa-tanahī jirm al-ʿālam, 207, where Kindī makes the claim that all composites (murakkabūna) have a composer (lahum murakkib), and that the two terms belong to the domain of relatives (bāb al-muḍāf). Translations of the relevant passage can be found in Adamson and Pormann, The Philosophical Works of al-Kindī, 67, and McGinnis and Reisman, Classical Arabic Philosophy, 23.

¹⁶For instance, not being a stone is an immediate inseparable accident of being alive. The technical term lāzim is also rendered as ‘implicate’.

¹⁷We would have to assume that (i) both ‘composite’ (muʾallaf) and ‘having a composer’ (annahū lahū muʾallif) follow immediately from the minor term ‘body’ (ğism) and (ii) that a demonstration is required because ‘composite’ is true evidently of ‘body’, and ‘having a composer’ is true evidently of ‘composite’ but not of ‘body’. This would exemplify the case Avicenna has discussed above and be consistent with the fact that neither ‘composite’ nor ‘having a composer’ are part of the definition of ‘body’ (which is discussed in Ilāhiyyāt 2.2 in terms of body's being a substance that can be divided along three dimensions). Problems remain, however, because the fact that body is a composite of matter and form seems to be something proved – with extensive arguments – in Ilāhiyyāt 2.2, and not something that follows immediately and evidently from the notion of body (see in particular Ilāhiyyāt 2.2, 73, where the claims that bodies are composed of matter and form seems to be presented as the outcome of the preceding discussion).

¹⁸فقد بان أن الحد الأكبر في الشيء المتيقن اليقين الحقيقي لا يجوز أن يكون علة للأوسط عسى أن يكون فيه جزء هو علة للحد الأوسط واعتبار الجزء غير اعتبار الكل فإن المؤلِّف شيء وذو المؤلِّف شيء آخر فإن ذا المؤلِّف هو بعينه محمول على المؤلَّف وأما المؤلِّف فمحال أن يكون محمولا على المؤلَّف .

¹⁹This is an admissible configuration, called sign (dalīl, which corresponds to the Aristotelian notion of sēmeion), but it is not immune from the possibility of doubt and therefore does not produce real certainty. Avicenna discusses dalīl in Burhān 1.7.

²⁰Reading muqawwim with Badawī’s edition (instead of maqūl in ʿAfīfī’s edition).

²¹لكن لقائل أن يقول إنه يجوز أن يكون الحد الأكبر غير مقوّم للأوسط بل هو أمر لازم له ومع ذلك ليس بمعلول له بل هو أمر مقارن له وكلاهما معا في الوجود ولكليهما علة في الوجود واحدة يشتركان فيها مثل الحال ين الأخ والأخ وكيف يمكننا أن نقول إن لزوم وجود الأخ عن الأخ إذ جعلناه حدا أوسط لزوم عن علة ومع ذلك فإنه يقيني لا شك فيه .

²²If the major is not a constituent of the middle, then it is not a cause, at least according to one way in which Avicenna understands causality: constituents are parts of the definition, so if we assume that the sort of cause Avicenna has in mind here is definition or part of a definition, then not being a constituent is a good proxy for something's not being a cause.

²³The condition of simultaneity for relative terms is already explicit in Aristotle's Categories, 7b15.

²⁴The implication (luzūm) associated with the correlation between two terms (taḍāyuf) is mentioned by Avicenna in his discussion of hypothetical conditional statements (šarṭiyyāt muttaṣila) in Qiyās 5.1, 233.17-234.9, translated in Shehaby, The Propositional Logic of Avicenna, 37. In the context of a passage where he is considering various kinds of necessary relationships between an antecedent (muqaddam) and a consequent (tālin), Avicenna lists correlatives (muḍāyif) among the types of terms that, in pair, may yield a necessary conditional (the others being cause-effect, effect-cause, each one effect of the cause of the other, each one effect of one and the same cause).

²⁵That is, it becomes certain only through the cause.

²⁶I take the pronoun in the expression min šaʾnihi to refer to aḥaduhumā at the beginning of the sentence.

²⁷أما إذا كان هاهنا أمران ليس أحدهما متعلقا بطبيعة الآخر بل تعلق أحدهما أو كلاهما بشيء آخر فإنه ليس أحدهما يجب بالآخر بل مع الآخر وإذا كان كذلك فليس أحدهما يتيقن بالآخر وأما إذا كان أحدهما علم من جهة العلة فإن كان الآخر علم أيضاً من جهة العلة فتوسيط الأمر الآخر لا يفيد يقينا بذاته إذ قد حصل ذلك من جهة العلة وأما إن كان أحدهما يعلم من جهة العلة والآخر مجهول لم يعلم بعلمه ثم من شأنه أن يعلم به الآخر فليس بينهما حال الإضافة فإن المضافين يحضران الذهن معا وإذا لم يكن كذلك لم يكن هذا جاريا مجرى الأخ والأخ إذ كان أحدهما أعرف للأصغر من الآخر.

²⁸For the distinction between symmetrical and asymmetrical relatives see Ilāhiyyāt, 3.10, 153, where the two types are referred to as equal (musāwin) or different (muḫtalif).

²⁹The truth of ‘Zayd has a brother’ is not certain because of the truth of ‘Zayd is a brother’. What grounds the certainty is a third thing, like being generated by the same parent. Similarly, the truth of ‘Zayd is a son’ is not certain because ‘Zayd has a father’ is true, but rather because there is an animal from whose sperm an animal of the same kind (Zayd) is begotten (see the passage from the Išārāt discussed below). Certainty about claims involving relative terms must be grounded in explanatory claims that do not involve relatives to avoid circularity.

³⁰The point made here echoes what Avicenna maintains in the passage quoted at the beginning of the paper, namely that failing to conceptualize ‘Zayd has a brother’ is equivalent to failing to conceptualize ‘Zayd is a brother’. The present argument runs as follows: relatives occur to the mind together; things that do not occur to the mind together are not relatives; A and B in this case do not occur to the mind together, because one is known through the cause, while the other is not; therefore, A and B are not relatives.

³¹Another context is Ilāhiyyāt 3.10, where Avicenna offers a detailed analysis of the category of relation and the relative, focusing in particular on the status of relations as accidents, the classifications of different types of relatives, the rejection of the infinite regress argument and the numerical distinctness of relations in different subjects. For an extensive discussion of these aspects, see Bäck, ‘Avicenna on Relation and the Bradleyan Regress’, 69–84; and Marmura, ‘Avicenna's Chapter “On the Relative”’, 83–99. In the context of Ilāhiyyāt 3.10, Avicenna discusses the example of brothers with a different purpose, namely to show that each one of two relatives each has a distinct relation towards the other: if A and B are brothers, then A's brotherhood in relation to B is distinct from B's brotherhood in relation to A (which Avicenna believes to hold, even more clearly, in the case of asymmetrical relatives like father and son).

³²The argument about relatives in Ilāhiyyāt 1.6 is in fact part of a larger argument aiming to show that the necessary existent cannot be equivalent (mutakāfī) nor relative (muḍāf) to something else. The first part of the argument focuses on the impossibility of there being two beings that are necessary in themselves and exist together without one being cause of the other.

³³والمضافان ليس أحدهما واجباً بالآخر بل مع الآخر والموجب لهما العلة التي جمعتهما وأيضاً المادتان والموضوعان أو الموصوفان بهما وليس يكفي وجود المادتين أو الموضوعين لهما وحدهما بل وجود ثالث يجمع بينهما.

³⁴Examples from the two lists of Ilāhiyyāt 3.10 include relatives like father and son, greater and lesser, double and half, brother and brother, tangent and tangent.

³⁵There seems to be a problem concerning the distinction between symmetrical relatives like brother and brother, and asymmetrical relatives like father and son. In the former case, a third distinct thing is certainly required to ground the relation. In the second case, it is unclear whether Avicenna thinks that the cause is one of the two (and therefore only the second is dependent on the first), or whether the genuine cause is something else. The first option seems to be supported by what he says in this passage, but there are other places (e.g. Ilāhiyyāt 6.2, 264-5, and a passage from Išārāt 1, 2.11 discussed below), where the father does not seem to be per se the cause of the son. For our purposes, however, in connection with Burhān 1.8, the example used here by Avicenna involves symmetrical relatives and the requirement of a cause, which is a third thing distinct from the two relatives, is not in question.

³⁶It is not clear why a necessary existent could not be a relative in this sense, given that it itself would be the cause. The argument probably requires that the two entities be equivalent, not just relative.

³⁷Avicenna seems to use a similar model also in Ilāhiyyāt 2.4, 81-2 while discussing the relation between matter and form.

³⁸A useful comparison may be made with the treatment of the relative in Avicenna's contemporary and rival Ibn al-Ṭayyib, see Ferrari, Der Kategorienkommentar von Abū l-Fara ğ ʿAbdallāh Ibn aṭ-Ṭayyib, esp. 285–99 (chapter 19) of the Arabic text.

³⁹This thought is also at the core of the passage quoted at the beginning of the paper, where Avicenna maintains that the statements ‘Zayd is a brother’ and ‘Zayd has a brother’ are equivalent in terms of conception (taṣawwur), as a result of which one either knows them both or fails to know them both at the same time, and proving one by means of the other is of no explanatory value.

⁴⁰Reading lahū instead of lahā.

⁴¹يجب أن يكون المعنى المعقول الذي للشىء الذى يحوج إلى أن يعقل معه غيره إنما هو له من أجل وجود ذلك الغير بإزائه فذلك المعنى الذى للشىء من أجل حصول الحال التى له ما صار الآخر معه هو إضافته مثل الأخ فإن حقيقة المفهوم من الأخوة لأحد الأخوين هو لأجل وجود الآخر وهى الحال التى له بسبب ذلك وهو كونه ابن أبى هذا الأول فإن الأخوة هى نفس اعتباره من حيث له آخر بهذه الصفة.

⁴²In this example the assumption is that A is the first son of a certain individual, say F, and B is the n-th son of F, where n > 1. On Avicenna's use of the male parent example see Bäck, ‘Avicenna on Relations’, 71, f.11. Avicenna's understanding of relations as being numerically distinct in different relatives, according to which for instance A's brotherhood with respect to B is numerically distinct from B's brotherhood with respect to A, is discussed in Ilāhiyyāt 3.10, 154–5.

⁴³See Avicenna, Nağāt, 174.6-11; translated in Ahmed, Avicenna's Deliverance: Logic, 139.

⁴⁴Graciousness: talaṭṭuf. Inati, Remarks and Admonitions, 75, understands the term in a similary way and translates it as ‘sensitive [answer]’; Goichon, Livre des directives et remarques, 110, takes a different line and renders it as ‘subtilité’.

⁴⁵قد يظن بعض الناس أنه لما كان المتضايفان يعلم كل واحد منهما مع الآخر أنه يجب من ذلك أن يعلم كل واحد منهما بالآخر فتؤخذ كل واحد منهما في تحديد الآخر جهلاً بالفرق بين ما لا يعلم الشيء إلا معه وبين ما لا يعلم الشيء إلا به وما لا يعلم الشيء إلا معه يكون لا محالة مجهولاً مع كون الشيء مجهولاً ومعلوماً مع كونه معلوماً وما لا يعلم الشيء إلا به يجب أن يكون معلوماً قبل الشيء لا مع الشيء ومن القبيح الفاحش أن يكون إنسان لا يعلم ما الإبن وما الأب فيسأل ما الأب فيقال هو الذي له ابن فيقول له لو كنت أعلم الإبن لما احتجت إلى استعلام الأب إذ كان العلم بهما معاً ليس الطريق هذا بل هاهنا ضرب آخر من التلطف مثل أن يقال مثلاً إن الأب حيوان يولّد آخراً من نوعه من نطفته من حيث هو كذلك فليس في جميع أجزاء هذا التبين شيء يتبين بالابن ولا فيه حوالة عليه .

⁴⁶ المتضايفان يكونان معا في الوجود والعقل فتعريف أحدهما بالآخر تعريف للشيء بالمساوي فيجب أن يعرّف كل واحد منهما بإيراد السبب الذي يقتضي كونهما متضايفين ليتحصلا منه معاً في العقل .

⁴⁷وهو الذي يضيف معنى الإضافة إلى الحيوان الذي هو الأب ويخص البيان به لأن الأب إنما يكون مضافاً إلى الإبن من هذه الحيثية.‏

⁴⁸Nearly half of Madḫal 1.9, the chapter on genus from Avicenna's section of the logic of the Šifāʾ that corresponds to Porphyry's Isagoge, is devoted to the discussion of correlatives along the lines of the Išārāt passage (the latter seems to draw almost verbatim on certain passages of the former, including the use of talaṭṭuf in Madḫal 1.9, 53.3). Most interestingly, however, instead of the father–son example, what we find in Madḫal 1.9, 53.4-7 is the brother case (one cannot define brother as the one that has a brother in a non-circular way), where the definition of the relative is given in causal terms, with reference to the cause of both, i.e. the father. This strategy, which is also reflected in the passage from Maqūlāt 4.3 presented above, confirms once more that the crucial aspect for the understanding of the relative case in Burhān 1.8 is the analysis of relatives in terms of their cause.